HSR&D Home » Research » IIR 10-374 – HSR&D Study
Patient-Level Determinants of Oral Anticoagulation Control in the VHA
Adam J Rose, MD MSc
VA Bedford HealthCare System, Bedford, MA
Funding Period: January 2011 - December 2013
Over 100,000 VHA patients receive oral anticoagulation therapy (OAT) each year. It is known that OAT is safer and more effective when patients spend more time in the therapeutic range (TTR). However, VHA sites of care vary widely on TTR, from 70% time in range (excellent control) to below 40% (poor control). Our previous efforts to adjust TTR for case mix have had certain limitations, and we need to know more about what sites might do to improve their TTR.
1) Improve our risk adjustment model for TTR. 2) Study the relationship between site-level processes of care and site-level performance on risk-adjusted TTR (RA-TTR). 3) Develop and pilot a tool to report each site's performance and opportunities for improvement.
We included all patients who received OAT from the VHA during the study period. We computed TTR for each patient and mean TTR for each site of care.
Aim 1) We improved our risk adjustment model by adding severity levels to comorbid conditions that impact TTR.
Aim 2) We studied the impact of four process of care measures on site-level anticoagulation control. Two of these measures relate to the site mean follow-up interval after a low (= 1.5) or high (>/= 4.0) INR (international normalized ratio) value. Another measure relates to the proportion of patients at each site with a mean INR value between 2.3-2.7, which is reflective of the use of a gudeline-concordant target INR range. The final measure relates to the rate of 56 day "gaps" in monitoring, or periods when patients are in possession of warfarin but not receiving INR testing.
Aim 3) We modified an existing SQL-based "Dashboard" to track these four process measures, and used it to profile site-level TTR and site-level process of care measures over time in VISN 1. End user feedback regarding the measurement system was solicited and incorporated.
With regard to the first aim, we finished our work in severity stratification for the most important clinical conditions that impact anticoagulation control. We examined how the severity of these conditions can impact anticoagulation control (TTR), major hemorrhage, and in some cases rates of ischemic stroke. Conditions we examined include chronic liver disease, dementia, mental illnesses, alcohol and non-alcohol substance abuse, congestive heart failure (CHF), and cancer.
We developed a composite severity score for CHF that works very well to risk-stratify patients on outcomes, based upon such factors as serum sodium, AST, a history of heart failure admissions, and use of certain medications (e.g., metolazone). Our composite severity score for patients with liver disease is based upon factors such as albumin level and creatinine level, which is important because little is currently understood about which patients with liver disease can safely receive warfarin. We noted that while patients with substance abuse have poor outcomes as a group, among patients with alcohol abuse, the AST/ALT ratio (aspartate aminotransferase / alanine aminotransferase) is extremely useful to risk-stratify with regard to predicting outcomes. In fact, patients with a normal AST/ALT ratio have similar outcomes to the general population of warfarin users, an important and reassuring finding. In general, most patients with mental illness have similar outcomes to the general population when using warfarin, especially after accounting for covariates.
With regard to the second aim, we developed four site-level process of care measures: mean follow-up interval after low INR; mean follow-up interval after high INR; rate of 56-day gaps in INR testing per patient-year; and proportion of patients with a mean INR value 2.3-2.7. Together, these 4 process measures have an R2 of 0.48 to predict site-level percent time in therapeutic range.
With regard to the third aim, we deployed these four process measures as part of a Dashboard in VISN 1 (Veterans Integrated Service Network 1). This Dashboard has been instrumental in helping to drive quality improvement in VISN 1 in this area. It has also served as a basis for another project to continue our QI work in anticoagulation.
Our 4 process of care measures for anticoagulation care, developed through this project, have become the basis of our effort to improve anticoagulation control in VISN 1 (SDP 12-249). Incorporating these measures into a SQL dashboard has enabled public reporting and ongoing quality improvement efforts. These measures have been well-received. Anticoagulation control (TTR) has already improved in VISN 1 from 63% to 66%, a meaningful change. This improvement has been correlated with improvements in the process of care measures. We are working to incorporate these process measures, developed through this project, as part of a national Anticoagulation Dashboard for the entire VHA.
Our work on risk-stratifying specific clinical conditions has contributed important insights that can help inform clinical practice.
Little had been known about warfarin use in patients who abuse alcohol, and clincians had been profoundly uncomfortable with the lack of information. We have identified the AST/ALT ratio as being able to risk-stratify these patients and to identify a sizeable subset whose outcomes are similar to the population at large.
Similarly, there has been no information about the use of warfarin in patients with liver disease. Our study (paper submitted) has demonstrated that two commonly measured lab tests (albumin and creatinine) can risk-stratify these patients, and can again identify a sizeable subset whose outcomes are similar to the population at large.
Our work on patients with mental illness has confirmed that most such patients can use warfarin with similar results to the general population.
Our work with CHF patients has demonstrated that easily available information can risk-stratify patients in a way that predicts TTR and also adverse events.
External Links for this Project
NIH ReporterGrant Number: I01HX000542-01
Dimensions for VADimensions for VA is a web-based tool available to VA staff that enables detailed searches of published research and research projects.
If you have VA-Intranet access, click here for more information vaww.hsrd.research.va.gov/dimensions/
VA staff not currently on the VA network can access Dimensions by registering for an account using their VA email address. Search Dimensions for this project
DRA: Cardiovascular Disease, Health Systems
DRE: Treatment - Observational
MeSH Terms: none